Research On Heart Rate Monitoring Method For Remote Rehabilitation Training Scene Video

Heart rate is an important physiological parameter.Accurate and effective monitoring of heart rate can not only reflect human status,but also prevent the occurrence of a series of diseases.When patients use a rehabilitation robot for rehabilitation training,real-time heart rate monitoring can reflect their physical condition during training.If the training intensity is excessive or they feel tired,they can be detected and terminated in a timely manner.Traditional heart rate measurement methods require a sensor to be directly connected to the skin when monitoring heart rate.When a patient is undergoing rehabilitation training,prolonged physical contact with the sensor may cause skin allergies and affect patient activity.In remote heart rate monitoring algorithms,factors such as the subject’s head movement and illumination can seriously affect the accuracy of the method’s heart rate estimation.To address these issues,this paper designed and implemented a facial region segmentation algorithm that integrates skin thickness and blood vessel distribution,and a chrominance(CHROM)video heart rate monitoring algorithm based on wavelet transform.The specific research content of this article is as follows:Firstly,aiming at the problem of region of interest division in video based heart rate monitoring algorithms,a facial region division algorithm integrating skin thickness and blood vessel distribution is proposed.This method combines computer vision with human physiological structures.Because the depth of blood vessels affects the diffuse reflection generated by blood,the face is divided into regions based on the thickness of the skin,and several candidate regions of interest are divided based on the distribution of facial blood vessels.Then,seven classical algorithms for remote heart rate monitoring are used to conduct experiments on candidate regions of interest.Finally,through comprehensive analysis of the performance indicators of each candidate region of interest for each method,excellent candidate regions of interest are obtained.Secondly,aiming at the problems of environmental noise and motion artifact interference,a remote heart rate monitoring algorithm based on wavelet transform for CHROM is proposed.This method uses two-dimensional discrete wavelet transform to perform multi-level decomposition of the video,and uses threshold technology to perform threshold denoising on the decomposed video frames,achieving filtering in both space and time.In order to solve the problem of motion artifacts,color difference signals are used to extract the heart rate source signal.After obtaining the heart rate source signal,the signal is converted to the frequency domain through Fast Fourier Transform and then the heart rate is estimated,resulting in a more accurate heart rate value.Finally,by integrating a facial region segmentation algorithm that integrates skin thickness and blood vessel distribution with a CHROM video heart rate detection algorithm based on wavelet transform,a CHROM heart rate monitoring algorithm based on wavelet transform under physiological characteristics is proposed.Experiments have verified the effectiveness of the algorithm on LGI data sets and self built data sets,further improving the accuracy and robustness of the algorithm.Finally,a remote monitoring system for the YSU-Ⅱ rehabilitation robot is designed and implemented for real-time monitoring of patients’ heart rate.